[H] XFX 290x DD Crossfire vs 780ti SLI Review

[Unoid]

source: http://hardocp.com/article/2014/01/26/xfx_r9_290x_double_dissipation_edition_crossfire_review/1

Looks aside which may very well mean nothing to you, these video cards well built and perform exceptionally. The video cards are sturdy, and very slim. You will be surprised how slim and thin these are as you install these in your system. The copper heatpipes remove heat faster from the core and the fans keep the heatsinks cool with super quietness. For the MSRP price of $569 these video cards are an appealing offer. We cannot wait to see what overclocking potential lies inside with the unlocked voltage.

my comments:
Interesting outcome. 290x's with proper cooling stock vs stock beat the 780Ti's. (4k/multiple monitor)
If/when you find these aftermarket cards priced near MSRP and don't wish to OC, crossfire is the way to go.

 

[Dman8777]

Case cooling will be more involved with the crossfire setup. 747 watts vs 522 for the nvidia system and all the heat stays in the case. Still, at the theoretical MSRP, the crossifre setup is definitely the better option.

 

[lavaheadache]

nice comparison. Too bad I swore off xfx sometime ago =P. There needs to be a few more releases like these cards to get me to give them a shot again.

Looking forward to the [H] follow up overclocking article. Ti's leave a lot of performance untapped where as my suspicion is that these 290x's are much more limited in what's left.

Still, color me impressed.

The only thing I have left to say is, Where is Mantle?

 

[Unoid]

Case cooling will be more involved with the crossfire setup. 747 watts vs 522 for the nvidia system and all the heat stays in the case. Still, at the theoretical MSRP, the crossifre setup is definitely the better option.

Kepler sure is more efficient power wise. Nvidia just isn't competitive on pricing unfortunately.

 

[Qbah]

Whoa... the Radeons destroy the 780Ti in SLI at 4K... there's like a 20%+ advantage across all the games! They do take 200W more at load though

I'm still perfectly happy with my 1080p and a single GTX670

 

[thilanliyan]

Wish the power consumption was lower, but besides that, XFX seems to have done a decent job this time, unlike the 7K DD series.

 

[Slomo4shO]

The only thing I have left to say is, Where is Mantle?

The beta drivers that support mantle are expected to drop this month. With the way DICE continues to fail with BF4, I am thinking there will be no real performance details until Thief is released on February 25.

 

[TreVader]

SLI looks like garbage in this review. 40% slower than CFX? I don't know why anybody would buy multi-gpu and go Nvidia after seeing this.

 

[lavaheadache]

SLI looks like garbage in this review. 40% slower than CFX? I don't know why anybody would buy multi-gpu and go Nvidia after seeing this.

lol, for years people said the same thing about Crossfire. These results aren't the norm nor are they set in stone as new drivers could easily swing things the other way.

 

[cmdrdredd]

Personally if I'm buying ~$600 cards I'm overclocking them.

 

[TreVader]

Personally if I'm buying ~$600 cards I'm overclocking them.

That really changes nothing. OC vs OC a 780ti is not gonna make up a 35% deficit to a 290x. It won't be AS BAD... but it still is gonna be 15-25% slower than AMD CFX


This is only if we're doing SLI vs CFX. Single GPU the 780ti is definitely the king.

 

[TreVader]

lol, for years people said the same thing about Crossfire. These results aren't the norm nor are they set in stone as new drivers could easily swing things the other way.

And for years Nvidia was considered the only way to go for mGPU. This isn't drivers, it's the 32 missing ROPs when you compare GK110 to Hawaii (SLI vs CFX)

 

[Unoid]

Don't forget that we don't have results at 1440p/1600p where a lot of enthusiasts like myself game. the 780Ti's may do better at the lower resolution.

 

[tolis626]

SLI looks like garbage in this review. 40% slower than CFX? I don't know why anybody would buy multi-gpu and go Nvidia after seeing this.

lol, for years people said the same thing about Crossfire. These results aren't the norm nor are they set in stone as new drivers could easily swing things the other way.

SLI never was garbage and probably never will.It's just that,for the first time in my rather short memory,I have to congratulate AMD for improving upon their existing technology that much.The latest Crossfire implementation is indeed really good,at least with regards to scaling and reliability.Does that mean NVidia suddenly is inferior?The answer is no.They've been superior in that aspect for far too long and it was about damn time that AMD caught up.Let's also not forget that the new CF is for the 290 series and beyond.If SLI wasn't that good,I doubt we would ever see any meaningful improvement to CF.

So,as it stands,right now CF is the better option for a new buyer.Things may change drastically in the next few months however.I don't expect NVidia sto stay still as AMD gets all the praise.

 

[Fallen Kell]

Let me know when AMD fixes the stuttering issues in crossfire....

 

[Gloomy]

Let me know when AMD fixes the stuttering issues in crossfire....

Stop trying to thread crap and read the review before posting.

Let’s talk about smoothness in gameplay in CrossFire versus SLI. With the AMD R9 290X series AMD fully supports Frame Pacing in Eyefinity and 4K. This option is enabled by default. Our feedback on this technology is that Frame Pacing works well to reduce choppiness and create a smooth and fluid gameplay experience with R9 290X CrossFire. We did not experience any choppiness or stuttering with R9 290X CrossFire like we do with the previous generation. There isn’t much to say on this topic anymore because with R9 290X CrossFire it just works, and it does a good job at it.

We only experienced one stuttering situation and that was with NVIDIA GeForce GTX 780 Ti SLI in Far Cry 3 at 4K. This could simply be a bug, but we did experience it, and it did affect the in-game settings we were able to play at. Overall, the smoothness between SLI and CrossFire in every game seems to be on par now. We have no major complaints.

 

[IEC]

Let me know when AMD fixes the stuttering issues in crossfire....

From the conclusion at [H]:

Frame Pacing

Let’s talk about smoothness in gameplay in CrossFire versus SLI. With the AMD R9 290X series AMD fully supports Frame Pacing in Eyefinity and 4K. This option is enabled by default. Our feedback on this technology is that Frame Pacing works well to reduce choppiness and create a smooth and fluid gameplay experience with R9 290X CrossFire. We did not experience any choppiness or stuttering with R9 290X CrossFire like we do with the previous generation. There isn’t much to say on this topic anymore because with R9 290X CrossFire it just works, and it does a good job at it.

We only experienced one stuttering situation and that was with NVIDIA GeForce GTX 780 Ti SLI in Far Cry 3 at 4K. This could simply be a bug, but we did experience it, and it did affect the in-game settings we were able to play at. Overall, the smoothness between SLI and CrossFire in every game seems to be on par now. We have no major complaints.

Mirrors my own experiences. I run 290 crossfire and it's the first trouble-free crossfire for me.

 

[SolMiester]

I remember on the 290 series release and 4k results, NV stated there were issues with 4k as no one had benched them before. Have NV done anything about that yet?

 

[Grooveriding]

Check out those Battlefield 4 minimums. For my uses 780ti SLI is unplayable at 4K or Eyefinity. Minimums in the 20s are terrible for a FPS. The 290X CF has 50% to 80% higher minimums.

780ti SLI is getting killed in the surround benches as well with 4xMSAA on. Wonder why nvidia performs so poorly with MSAA in BF4 ? I see that in benches of the game all the time, nvidia just getting hammered by MSAA.

 

[Gloomy]

Kepler sure is more efficient power wise. Nvidia just isn't competitive on pricing unfortunately.

I don't trust their "Full Load" testing. All it takes for power consumption to be skewed is a bad SLI or CF profile for the game they use. Since SLI isn't performing at all in this review it's safe to say that it's not scaling, or not a full load, so it's not representative of power consumption either.

In their 780Ti review they hit 410w full load with a single card. This means that they somehow sourced a magical 112w 780Ti as the second card for this review.

780Ti SLI should be in the ballpark of 620w. I think it may be a typo...

 

[Will Robinson]

That is a sweet result for AMD R9.
Interesting to see the fabled 780Ti get soundly beaten in XFire V SLI.
They're a lot more expensive as well aren't they?
Its a wonder this thread doesn't have 5 pages by now.^_^

 

[omeds]

290x with 64 ROP's and uber fill rates simply dominates GK110 at eyefinity/4k resolutions, despite GK110 chips being faster when not fill rate limited at lower resolutions (1080p/1440p). Basically, we are witnessing GK110 choke or become limited a lot earlier as resolution increases.

 

[cmdrdredd]

290x with 64 ROP's and uber fill rates simply dominates GK110 at eyefinity/4k resolutions, despite GK110 chips being faster when not fill rate limited at lower resolutions (1080p/1440p). Basically, we are witnessing GK110 choke or become limited a lot earlier as resolution increases.

Or some type of driver issue with SLI.

 

[el etro]

Could be the 3GB ram wall been hit at 5760*1080 or 4k??

 

[omeds]

Or some type of driver issue with SLI.

Possibly, but it's no secret Hawaii has the advantage over GK110 when they become fill rate bound at higher resolutions.
To be fair, Kepler is old, it's real competition was Tahiti.

Could be the 3GB ram wall been hit at 5760*1080 or 4k??

On occasions I'm sure that was the case too.